Closed Bug 763982 Opened 8 years ago Closed 8 years ago

Arabic-character processing in nsTextFrameUtils::TransformText is out of date

Categories

(Core :: Layout: Text and Fonts, defect)

defect
Not set

Tracking

()

RESOLVED FIXED
mozilla16

People

(Reporter: jfkthame, Assigned: jfkthame)

Details

(Keywords: intl)

Attachments

(1 file)

nsTextFrameUtils::TransformText is responsible for detecting Arabic characters, in order to support the bidi.numeral processing within textruns. However, the macros it uses to identify Arabic characters are out of date with respect to Unicode (they don't account for the Arabic Supplement and Extension blocks). They can also be rewritten slightly to run more efficiently, which is important because TransformText is showing up as a significant contributor to profiles in bug 762710, for example.

The attached patch updates the Arabic character range, and by optimizing the order of comparisons, it also shaves around 5-10ms off the reflow time for the tinderbox log in bug 762710#c0 (on my MBP).
Attachment #632276 - Flags: review?(smontagu)
Attachment #632276 - Flags: review?(smontagu) → review+
https://hg.mozilla.org/integration/mozilla-inbound/rev/42e5e3988682
Assignee: nobody → jfkthame
Target Milestone: --- → mozilla16
https://hg.mozilla.org/mozilla-central/rev/42e5e3988682
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.